A network approach reveals roles for cancer stem cells in cancer promotion

Mark: I’m from the South in the United States.  I didn’t know what to do with my life (still don’t) but I liked biology, so got a PhD after college and really liked evolutionary genetics and the people who do it! I chose to go to medical school at UCSF because the research is so awesome.  I loved Allan’s conceptually integrated and driven approaches.  Single-cell cancer models are an evolutionary biologist’s dream – repeated evolutionary outcomes (tumours) with tons of genotypic, phenotypic, and fitness data.  With this system, we can really ask when we replay the tape of life, does it play out in the same way?

Mark: The main advances underlying this project were that human cancer has been shown to be dependent on exposure to environmental carcinogens, but is also strongly affected by individual genetic background. We wanted to replicate these factors in mouse cancer models, as most genetic models of cancer in mice are based on one mouse strain, and use massive and unrealistic genetic mutations, that are rarely seen in the real world, to induce cancers. Here, we introduced genetic diversity and exposure to mutagenic agents, as well as promoting factors, to generate a more realistic cancer model that reflects the human cancer risk factors. 

Allan: This paper addresses the controversies surrounding the roles of so-called “stem cells” in cancer development. The controversy arises from the fact that in solid epithelial tissues such as the skin there are many different stem cell populations, several of which have been proposed to be “cancer stem cells” capable of driving cancer development. At least part of this controversy arises from the methods previously used to define stem cells and measure their functions. When we look at the expression of different individual stem cell genes, they are often expressed in different cells, in normal tissue or in tumours, with no obvious overlaps. The manuscript describes a way to construct an expression “network” for stem cell marker genes, and to combine this with single cell RNASeq to look at the expression of each network in single cells, rather than just the single stem cell genes. 

Allan: This approach shows that in contrast to expression of single stem cell genes, the expression of their networks overlap significantly, especially in cells that are highly stressed during the rapid growth of pre-neoplasia. These networks represent a high plasticity stem cell state capable of responding to different kinds of stresses during growth. 

Mark: Gene networks generated from hundreds of bulk samples reveal conserved patterns of transcriptomic dysregulation in single cells of skin cancer.  These patterns are particularly important in early neoplasia before florid malignancy is established.  In fact, they revealed that there were two main cell states in pre-neoplastic cells, one marked by stemness and the other by proliferation.  The toggling between these states may control progression to from benign to malignant tumours.

Allan: The model proposed is that constant exposure to tumour promoters causes rapid early clonal expansion, which induces several kinds of cellular stress including oxidative stress, endoplasmic reticulum stress, tumour suppressor stress, and immune attack, causing a variety of stem cell networks to be activated in the same cell state. This high plasticity cell state can be resolved by genetic and epigenetic changes leading to proliferation and expression of a different set of networks.

An important conclusion from the paper is that the high plasticity stem cell state seen in these early pre-malignant tumours is also induced when fully malignant tumours are treated with chemotherapy. Mouse tumour cells stressed by cis-platin re-express the same stem cell network state found in pre-neoplasia, and this also applies to several types of human cancers in patients treated with different cancer drugs. All seem to converge on the same network state which has been linked to drug resistance. Manipulating these states can help us to find new combinations of treatments for different cancer types.

Mark: For me, the real surprise of this project came gradually as the conserved patterns began to emerge.  At first, it seemed too good to be true.  We kept seeing the same cell clusters expressing particular stem networks in a particular way, but thought it was a fluke at first.  But after a lot of attempting to figure out why, we decided that these patterns are probably robust and a real feature of both mouse and human cancer.

Mark: Understanding how broadly conserved these patterns really are.  Many other groups have identified similarly simplified patterns, but in different ways.  Uniting the field’s understanding of these states, and understanding whether they’re meaningfully distinct in different cancers, is a major next step.

 "If we can understand these stem cell states, we can possibly manipulate them to make cancers more susceptible to treatment."

Dr Mark Taylor

Allan is also a member of Cancer Grand Challenges team Mutographs. Work from Allan’s lab showing that many carcinogens do not directly cause mutations, together with other findings from team Mutographs, caused a revival in thinking around the promotional hypothesis of carcinogenesis and inspired the Normal Phenotypes challenge. Team PROMINENT are delving deeper into the potential mechanisms 'promoters' use to stimulate neoplastic growth. They are funded by Cancer Research UK, the National Cancer Institute in the US and Fundación Científica de la Asociación Española Contra el Cáncer. Find out more about PROMINENT.